Developing a Flavor Wheel Using Machine Learning
Author Information
Author(s): Rodríguez-Mendoza Anggie V., Arbeláez-Parra Santiago, Amaya-Gómez Rafael, Ratkovich Nicolas
Primary Institution: Universidad de los Andes
Hypothesis
This study aims to investigate the relationships between chemical compounds and sensory descriptors in distilled spirits using machine learning techniques.
Conclusion
The study successfully developed a flavor wheel that categorizes and interrelates descriptors of various distilled spirits, enhancing understanding and appreciation of their complexities.
Supporting Evidence
- The study analyzed a comprehensive dataset with 3051 chemical compounds.
- Principal component analysis was used to reduce the dimensionality of the dataset.
- A clustering machine learning model identified distinct clusters of aroma descriptors.
- The developed flavor wheel serves as a valuable tool for distillers and consumers.
- The research highlights the intricate relationships between chemical compounds and sensory descriptors.
Takeaway
The researchers used computers to help create a colorful chart that shows different smells and tastes in drinks like whiskey and rum, making it easier for people to understand what they are tasting.
Methodology
The study employed principal component analysis (PCA) and clustering techniques to analyze a dataset of 3051 chemical compounds and their associated aroma descriptors for seven distilled spirit categories.
Potential Biases
The descriptors used to create the flavor wheel may have been influenced by the sensory profiles of particular types of beverages.
Limitations
The flavor wheel may not cover the full spectrum of sensory experiences and could be biased towards specific distilled beverages.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website